-
Notifications
You must be signed in to change notification settings - Fork 7
/
perceptron_tagger.py
190 lines (178 loc) · 6.74 KB
/
perceptron_tagger.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
import pickle
import collections
from averaged_perceptron import AveragedPerceptron
import random
import logging
logging.basicConfig(level=logging.INFO)
def _pc(c,n):
return float(round(c/n,6))*100
class PerceptronTagger(object):
_START = ['-START2-','-START-']
_END = ['-END-','-END2-']
def __init__(self,bool_load):
self.classes = set()
self.tagdict = dict()
self.model = AveragedPerceptron()
if bool_load:
self.load(PICKLE)
def _get_features(self, i, word, context, prev, prev2):
features = collections.defaultdict(int)
i += len(self._START)
def add(name, *args):
features[' '.join((name,) + tuple(args))] += 1
add('bias') ##意义在于哪？？？
add('i suffix ', word[-3:])
try:
add('i prev1 ', word[0])
except IndexError as e:
print(e.args)
add('i-1 tag ', prev)
add('i-2 tag ', prev2)
add('i tag+i-2 tag ', prev, prev2) # flag 有问题
add('i word ', context[i])
add('i-1 tag+i word ', prev, context[i])
add('i-1 word ', context[i - 1])
add('i-1 suffix ', context[i - 1][-3:])
add('i-2 word ', context[i - 2])
add('i+1 word ', context[i + 1])
add('i+1 suffix ', context[i + 1][-3:])
add('i+2 word ', context[i + 2])
return features
def _normalize(self, word):
if '-' in word and word[0] != '-':
return '!HYPHEN'
elif word.isdigit() and len(word) == 4:
return '!YEAR'
elif word.isdigit():
return '!DIGIT'
else:
return word.lower()
def _make_tagdict(self, sentences):
freq_thres = 20
ambiguity_thres = 0.97
counts = collections.defaultdict(lambda: collections.defaultdict(int))
for words,tags in sentences:
for tag, word in zip(tags, words):
counts[word][tag] += 1
self.classes.add(tag)
for word, tags in counts.items():
n = sum(tags.values())
tag,mode=max(tags.items(),key=lambda x:x[1])
# for tag, num in tags.items():
if n > freq_thres and float(mode) / n > ambiguity_thres:
self.tagdict[word] = tag
def load(self, loc):
try:
w_td_c = pickle.load(open(loc, 'rb'))
except IOError as e:
msg = ("Missing trontagger.pickle file.")
raise IOError(msg)
self.model.weights,self.classes,self.tagdict = w_td_c
self.model.set_classes(self.classes)
def tag(self,corpus):
s_split=lambda x:x.split('\n')
w_split=lambda x:x.split()
def split_corpus(corpus):
for sentence in s_split(corpus):
yield w_split(sentence)
tokens=list()
prev,prev2=self._START
for sentence in split_corpus(corpus):
context=self._START+[self._normalize(word) for word in sentence]+self._END
for index,word in enumerate(sentence):
# word_normalize=self._normalize(word)
word_normalize=word
guess=self.tagdict.get(word_normalize)
if not guess:
features=self._get_features(index,word,context,prev,prev2)
guess=self.model.predict(features)
tokens.append((word,guess))
prev2=prev
prev=guess
return tokens
def train(self, sentences, save_loc=None, nr_iter=5):
logging.info('Start training.....')
self._make_tagdict(sentences)
# print('self.tagdict==>\n',self.tagdict)
self.model.set_classes(self.classes)
for iter_n in range(nr_iter):
c,n=0,0
for words,tags in sentences:
prev,prev2=self._START
# context=self._START+[self._normalize(word) for word in words]+self._END
context=self._START+[word for word in words]+self._END
for index,word in enumerate(words):
features=self._get_features(index,word,context,prev,prev2)
guess=self.tagdict.get(word)
if not guess:
guess=self.model.predict(features)
self.model.update(guess,tags[index],features)
c+=guess==tags[index]
n+=1
prev2=prev
prev=guess
# logging.info("Iter {0}: {1}/{2}={3}".format(iter_, c, n, _pc(c, n)))
logging.info('Iter {0}: {1}/{2}={3}'.format(iter_n,c,n,_pc(c,n)))
random.shuffle(sentences)
self.model.average_weights()
if save_loc:
pickle.dump([self.model.weights,self.model.get_classes,self.tagdict],open(save_loc,'wb'))
else:
pass
return None
if __name__=='__main__':
pt=PerceptronTagger(False)
PICKLE = "data/trontagger-0.1.0.pickle"
train_filepath='data/train.txt'
test_filepath='data/test.txt'
nr_iter=5
try:
pt.load(PICKLE)
logging.info('Start Testing....')
fr=open(test_filepath,'r')
sentence=[[],[]]
right,total=0,0
for line in fr:
words = line.split()
# param=(words[0],words[1])
# sentence.append(param)
if len(words) != 2:
continue
sentence[0].append(words[0])
sentence[1].append(words[1])
if words[0] == '.':
text=''
for i, word in enumerate(sentence[0]):
text += word
if i < len(sentence[0]):
text += ' '
tokens=pt.tag(text)
for index,(word_iter,guess_iter) in enumerate(tokens):
try:
if sentence[1][index]==guess_iter.strip('\r\n '):
right+=1
except IndexError as e:
print(e.args)
total+=1
sentence = [[], []]
else:
pass
logging.info("Precision : %f", right / total)
except IOError as e:
training_data=list()
fr=open(train_filepath,'r')
sentence=[[],[]]
logging.info('Reading Corpus .....')
for line in fr:
words=line.split('\t')
# param=(words[0],words[1])
# sentence.append(param)
sentence[0].append(words[0])
sentence[1].append(words[1])
if words[0]=='.':
training_data.append(sentence)
sentence = [[], []]
else:
pass
logging.info('training corpus size {}'.format(len(training_data)))
pt.train(training_data,PICKLE,nr_iter)